adselect.stats package

adselect.stats.cache module

adselect.stats.cache.add_keyword_banner(publisher_id, banner_size, keyword, keyword_score, banner_id, limit=100)

Add keywords to cache per banner per publisher

Parameters:

• publisher_id –
• banner_size –
• keyword –
• keyword_score –
• banner_id –
• limit –

Returns:

adselect.stats.cache.delete_impression_count(banner_id)

Removes banner impression count.

Parameters:banner_id – Banner identifier. Returns:

adselect.stats.cache.reset_keyword_banners()

Reset the KEYWORDS_BANNERS cache to empty.

Returns:None

adselect.stats.tasks module

adselect.stats.tasks.calculate_last_round_score(*args, **kwargs)

Parameters:

• publisher_id –
• banner_id –
• keyword –

Returns:

adselect.stats.tasks.clean_database(*args, **kwargs)

Remove finished campaigns and associated stats.

Returns:

adselect.stats.tasks.configure_tasks()

Recalculate stats.

Returns:

adselect.stats.tasks.recalculate_stats(*args, **kwargs)

Dump all data from cache to database, reload cache data and recalculate everything.

Returns:

adselect.stats.tasks.save_banner_scores(*args, **kwargs)

Save scores. 1. Get scores from database. 2. Validate active banners. 3. For each banner get impression count

Returns:

adselect.stats.tasks.save_impression_count(*args, **kwargs)

Save impression count data from cache to database.

Returns:

adselect.stats.tasks.save_keyword_payments(*args, **kwargs)

Save payment per keywords data from cache to database.

Returns:

adselect.stats.tasks.save_new_banner_scores(*args, **kwargs)

Save scores for new banners

Parameters:db_banners – Set of banners already scored. Returns:

adselect.stats.utils module

adselect.stats.utils.genkey(key, val, delimiter='_')

Generate keyword identifier, ex. {‘animal’: ‘dog’} becomes ‘animal_dog’

Parameters:

• key – Key
• val – Value
• delimiter – Delimiter, default “_”

Returns:

Generated identifier

adselect.stats.utils.get_banners_for_keywords(publisher_id, banner_size, sbest_pi_keys, banners_per_keyword_cutoff=10)

Get publisher banners and get only best ones for keywords, with included cutoff.

Parameters:

• publisher_id – Id of the publisher
• banner_size – Banner size
• sbest_pi_keys – Best paid keywords
• banners_per_keyword_cutoff – Number of banners returned

Returns:

adselect.stats.utils.initialize_stats(*args, **kwargs)

Initialize data cache. Load data from the database into memory.

Returns:

adselect.stats.utils.is_banner_active(*args, **kwargs)

Check if banner is in the database, together with the campaign and if the campaign is active.

Parameters:banner – Banner id or banner document. Returns:True or False

adselect.stats.utils.is_campaign_active(campaign_doc)

Compare campaign’s start and end times with current time.

Parameters:campaign_doc – Campaign document. Returns:True for active campaigns, False for inactive.

adselect.stats.utils.iterate_deferred(*args, **kwargs)

Auxiliary function to iterate a function over a deferred resource.

Parameters:

• deferred – Deferred we iteravet over
• func – Function executed for each item

Returns:

None

adselect.stats.utils.load_banners(*args, **kwargs)

Load only active banners to cache.

adselect.stats.utils.load_impression_counts(*args, **kwargs)

Load impressions/events counts to cache.

adselect.stats.utils.load_scores(*args, **kwargs)

Load best paid keywords taking into account scores.

1. Get banner scores.
2. For each one, get banner.
3. Get

Parameters:scores_db_stats –

adselect.stats.utils.mixin_new_banners(selected_banners, propositions_nb, new_banners)

Add new banners without payment statistic

Parameters:

• selected_banners – Pre-selected banners
• propositions_nb – Amount of banners returned
• new_banners – Banners with amount of payments below threshold

Returns:

adselect.stats.utils.process_impression(banner_id, publisher_id, impression_keywords, paid_amount, increment=True)

Update impression cache.

1. Increase impression count.
2. If paid (paid > 0), update keyword paid amount.

Parameters:

• banner_id – Banner identifier.
• publisher_id – Publisher identifier.
• impression_keywords – Dictionary of keywords (with values).
• paid_amount – Amount paid for the impression.
• increment – Increment views

Returns:

adselect.stats.utils.select_best_banners(publisher_id, banner_size, sbest_pi_keys)

Select banners with appropriate size for given keywords.

1. Get best paid keywords (limited to a cutoff value)
2. Find common keywords from given keywords and best paid keywords.
3. Find best paid banners for give keywords and size.
4. Add new banners, which have no payments statistic yet.
5. Shuffle the banners.
6. Return a list of banners, size limited to the defined cutoff value.

Parameters:

• publisher_id – Publisher identifier.
• banner_size – Banner size (width x height) in string format.
• sbest_pi_keys – Best Keywords for this impression and publisher

Returns:

List of banners.

adselect.stats.utils.select_best_keywords(publisher_id, banner_size, impression_keywords_dict, best_keywords_cutoff=100)

Parameters:

• publisher_id – Publisher identifier.
• banner_size – Banner size (width x height) in string format.
• impression_keywords_dict – Dictionary of keywords for the request.
• best_keywords_cutoff – Cutoff of the number of best paid keywords taking into account.

Returns:

adselect.stats.utils.select_new_banners(publisher_id, banner_size, new_banners_proposition_nb)

Return banners ids without payment statistic.

The function doesn’t allow to display banners more than notpaid_display_cutoff times without payment.

1. Get banners with the right size.
2. Choose random banners from that population.
3. Filter out banners which were displayed less times than notpaid_display_cutoff
4. Return chosen banners.

Parameters:

• publisher_id – Publisher identifier.
• banner_size – Banner size (width x height) in string format.
• new_banners_proposition_nb – The max amount of new banners.

Returns:

List of banners.